THE CALIBRATION OF SEGMENTED GAMMA SCANNERS USING ROD SOURCES

ABSTRACT

The Segmented Gamma Scanner (SGS) is a widely applied instrument for the assay of suitable radionuclides in low-density radioactive waste drums. The approach involves an axial scan of the item as it is rotated, using a collimated high-resolution gamma spectrometer. A transmission source is frequently employed to determine the energy dependent transmission factor across each layer.

The calibration of the SGS is based on the assumption of a uniform radionuclide distribution, along with matrix homogeneity across each segment. To reduce the time and effort involved in experimentally determining the calibration parameters, a common practice is to simulate uniform source/matrix conditions by inserting an array of 6 or 7 mixed nuclide rod sources into channels placed in simple materials that cover the expected density range.

Errors introduced by the approximation of the uniform source distribution by a fixed pattern of multiple ideal line sources are examined. We show that placing the rods at the centroid of equal volume cylindrical shells provides an excellent analog. Additional factors such as the net energy dependent effect on the apparent specific emission rate at the detector due to the finite self attenuation suffered in the structure of the rods, including the associated re-entrant channels, and the displaced matrix material are considered. We present a table of correction factors for common nuclides and test matrices. It is concluded that the multi rod approach can be used to high accuracy over an extremely broad range of transmission factors.

INTRODUCTION

Transmission corrected Segmented Gamma Scanning is a well-known and widely applied method for the qualification of g-emitters in waste drums [1, 2]. Several hundred Segmented Gamma Scanners are currently in routine use throughout the world making them one of the most popular non-destructive assay tools. The SGS technique combines quantitative high-resolution g–spectrometry with an axial scan of the item as it is rotated to produce an estimate of the activity segment by segment. Matrix attenuation corrections, for each segment, are derived from diametrical transmission factors measured using an external multi-energy transmission beam. The fundamental assumptions underpinning the application of the SGS are that within each segment the matrix material and density is constant and homogeneous and furthermore that the distribution of activity is uniform in the matrix. In practice neither of these assumptions is satisfied perfectly. This is why the drum is continuously rotated for an integral number of rotations per segment during the counting period. During the Transmission measurement the rotation results in an approximate averaging of matrix inhomogeneities while in Emission mode it approximately evens out inverse square and attenuation variations due to a non-uniform activity distribution. As a matter of pragmatism the choice of scanning pattern is a trade off between accuracy and precision. Performing a scan with high spatial definition in principle may result in a more accurate assay result by adhering more closely to the underlying assumptions but it will also take longer so that given time constraints detection limits or throughput performance will be impaired. Because of this, typically only between 8-16 segments are used and a slot collimator some 4” tall by 10” wide by 8” long located perhaps only 4” from the drum wall would be a reasonable compromise. Because such an arrangement defines a broad field of view extending to segments above and below the one nominally selected during the axial scan, there is also an implicit assumption that the axial variation of the waste is smooth and gradual. Recall too that the Transmission beam may only be a fraction of the height of the segments being used, so that the axial coverage, even within a segment, is limited.

Given the fundamental assumptions of the SGS technique and the general method of practical implementation, the method is most reliable when applied to items with homogeneous matrix and uniform activity distributions, or, to items of low-to-medium gross density ( £0.5 g.cm ^-3, say in the case of 200L drum; reference [1] covers diametrical transmission factors of 1% or greater) known to have modest variation or alternatively known to have sufficient randomness that the averaging processes introduced by the scanning procedure are effective.

In practice there is ample empirical evidence that the SGS approach works remarkably well across a wide range of assay conditions and is sufficient to meet a broad range of regulatory and operational needs. Solid items, say, represent a severe case of inhomogeneity but are often only contaminated on the surfaces so that frequently this kind of violation of the SGS assumptions again gets smoothed out. The net result is that provided there are at least a few random ‘pockets’ (or ‘hot regions’) of activity in the drum, perhaps as few as 3-5, the SGS approach works well.

In accordance with the underlying assumptions and operational experiences it is, therefore, conventional to calibrate SGS instruments under conditions of homogeneous matrix and uniform activity. Allowance for ‘point-source’ effects is then accommodated by the Total Measurement Uncertainty. To reduce the time and effort involved in experimentally determining the calibration parameters, a common practice is to simulate uniform conditions by inserting an array of 6 or 7 mixed nuclide rod sources into channels placed in simple, often stable and self-supporting, homogeneous materials that cover the expected density range. The use of rods simulates a uniform axial distribution rather well. The radial disposition of the rods is chosen to approximate a uniform radial distribution (in terms of per unit volume).

In this paper we address the following questions:

1. how well does a fixed pattern of 6 or 7 ideal line sources mimic a uniform activity distribution?

and

2. what is the net energy dependent effect on the apparent specific emission rate at the detector due to the finite self attenuation suffered in the structure of the rods, including the associated re-entrant channels, and the displaced matrix material?

This is all part of ensuring that the calibration is accurate and well understood. It is important to bear in mind, however, that the uncertainties associated with measuring actual waste drums may, of course, be much larger than the uncertainty associated with the calibration procedure. The matrix and activity distributions will seldom be ideal. Even the variation in fill height introduces end effects bias. For the purposes of discussion we describe a particular calibration geometry although the method is quite general.